“Those who cannot remember the past are condemned to repeat it.” George Santa
It’s been 14 years since the mortgage crash (anyone remember MortgageImplode.com)? And in that time a whole generation of underwriters and loan officers have come into the business and/or moved up in the ranks of their companies. What they’ve experienced to date is an origination market that was heavily refinance-driven and relatively free of fraud. But that market is rapidly changing, and with it the risk of fraud is rising.
What’s changed of course is the overall size of the market, which is expected to contract by 40% this year, as well as its make-up, which has rapidly shifted from refinance to purchase. These dynamics are important for two reasons: first, fraud tends to be more prevalent in a purchase market environment and second, excess capacity often leads to cutting corners and “massaging” numbers to close deals.
There are other factors to consider as well. Dramatic home price appreciation is fostering FOMO—fear of missing out—among buyers and fix-and-flip investors. Meanwhile, rapidly climbing interest rates and loan-level price adjustments on non-owner-occupied properties are making properties less affordable. Historically, these conditions have fostered fraud. For example, the last time housing affordability was this low occurred in 2006 – the boom year that preceded the mortgage crisis, and a period during which mortgage fraud was pervasive.
Different Forms of Fraud
Mortgage fraud generally falls into two categories: fraud for housing and fraud for profit. Fraud for housing is the more common of the two, and it is relatively easy to commit. It occurs when a borrower misrepresents or omits relevant details about employment and income, debt and credit, or property specifics to obtain a home or an investment property. Sometimes a loan officer is complicit in these misrepresentations. Overstated income and undisclosed debt are the most prevalent types of fraud for housing. But falsified documentation, straw buyers and investment income misrepresentation can also come into play. Claiming that a second home or an investment property is really a primary residence also falls under this category, and higher interest rates for these kinds or properties, combined with new LLPAs from the GSEs, may trigger more of these misrepresentations.
Fraud-for-profit schemes are less common because they typically involve multiple participants in the mortgage process: mortgage brokers, appraisers, loan officers and/or straw buyers. While they can include borrower misrepresentations, they often also involve schemes to inflate (or sometimes deflate) the value of a property to ultimately defraud the lender or the borrower.
Why has Fraud Been on the Retreat?
Why has fraud been so low for the past decade, at least in the mortgage space? Better products (no subprime or “liar” loans), tighter underwriting (ability-to-repay requirements) and wider use of fraud detection solutions, such as the market-leading First American FraudGuard® solution, are all major factors.
Originally created as a first-generation fraud prevention and detection tool, FraudGuard has, over time, evolved to also cover loan quality, data integrity, fraud prevention and compliance, making it a go-to solution for overall QC and risk mitigation. Underwriters rely on FraudGuard to “fire” critical and high-risk alerts with accompanying reason codes for the variances. The solution tracks the steps that the underwriter has taken to clear the alert and captures this information for the lender and future investors.
Depending on a lender’s business rules, FraudGuard can detect more than 200 different variants in a loan application, many of which are due to data omission errors and data validation errors, as opposed to fraud. While customization can, and does, reduce the frequency of alerts, there is always a lingering concern among lenders that arbitrarily suppressing alerts will allow some risky loans to slip through.
At the same time, false positives can create unnecessary reviews of loans that, in reality, require a very low level of diligence related to risk. These false positives can ultimately slow the approval process and create trust and adoption issues with users. Some industry fraud experts are now concerned that a combination of a new generation of underwriters who grew up in a low fraud environment and a perception that many alerts are false positives might prompt some firms to lower their guard and expose themselves to more fraud.
Marrying the Best of Two Approaches
Last year First American and a large multi-channel lender client that had been using FraudGuard as its primary fraud/workflow tool embarked on an extensive test to determine if combining the alert-based tool with new AI fraud-pattern recognition technology would safely reduce the number of alerts that needed to be reviewed.
A prototype of what is now the First American AppIntelligence® Score and FraudGuard were used to review more than 25,000 live loans over an eight-month period. All of the loans were reviewed twice: once to identify loans that would have had to be reviewed by an underwriter because of either a fraud indicator or a loan quality defect; and then the alerts themselves were reviewed to see how the two aligned.
By layering the AppIntelligence Score technology with FraudGuard, the client was able to reduce loan reviews by more than 50% and eliminate alerts on low-risk issues by more than 40%. As a result, more loans could be handled via automation, and underwriters could focus their attention and efforts on the riskier loans and more critical alerts.
Based on the initial success of this combined solution, several other large clients are now in the process of conducting similar exercises.